[[alternative]]A STUDY ON THE FINANCIAL CRISIS PREDICTIVE MODEL OF THE LISTED ELECTRONIC COMPANIES

[[abstract]] 當一家公司發生財務危機，它不僅是傷害企業本身還會影響相關企業及其債權人或股東，特別是在資訊非常發達之今日社會，相關之危機消息將造成股票市場之信心崩潰，影響層面非大。因此，當公司發生財務危機時如何管理，以及在平時如何設置早期預警系統，以防止危機發生同樣是非常重要。 本研究乃嘗試用財務變數來討論和預測財務危機，而以台灣證券交易所於2000年至2003年公開發佈之上市上櫃電子公司財務報表所計算之財務變數為基礎，經蒐集十六家財務危機公司及七十二家財務正常公司合計八十八家公司為樣本，以每年之財務變數來探討財務危機發生之模型並計算其正確率，這些變數包括管理能力、償債能力、獲利能力、成長能力及財務結構等五項構面共二十七項自變數，而以公司是否發生財務危機為依變數，將這些變數經蒐集後以迴歸方法分析之。 結依迴歸分析方法中之反向淘汰法及逐步迴歸分析法分析結果獲得下列之結論：1、 依反向淘汰法迴歸分析獲得之結果，最後一年之正確率為93%，第二年之正確率為86%，第三年之正確率為90%，第四年之正確率為83%。2、 依逐步迴歸分析法驗證之結果，最後一年之正確率為91%，第二年之正確率為86%，第三年之正確率為83%，第四年之正確率為82%。3、 以89年至92年之迴歸分析建立之預警模型，檢驗該期間之財務狀況，其準確率達到91%。又以九十二年之財務比率，套用89年至90年之迴歸分析建立之預警模型，其準確率達到84%。4、 再以鑑別分析法對92年度之財務狀況重複驗證，其準確率達到84%。 本研究所使用之預警模型對於投資人、金融機構、保險業等於投資或授信時應具有一定程度之參考之價值，然而美中不止者乃是財務報表被窗飾情形仍相當嚴重，為保障投資人、金融機構或保險業等之投資、授信風險，建請有關單位加強財務報表之稽核。[[abstract]] When companies occurs financial crisis, it not only harms to the own business but also affect the related business or investors and stockholders. Especially in such days of wide-open information, the related crisis news worse results such like the broken-confident to the whole market. Therefore, to control the crisis companies well while the crisis is occurred is necessary, and to set up early warming system to avoid the happening of crisis in usual time is quietly important too. The research tries to use the financial variable to discuss and to predict the business crisis which is according to the companies in public that suffering to be the list for full delivery stock from Taiwan stock exchange market, or temporary ceasing operation stock as the samples during 2000 to 2003. 16 crisis companies sample is collected while the total company sample reaches 88. By the pre-four years financial variable before the moment year of crisis happening can create the model and count the accurate of the crisis. These variables including management capability, pay debt capability, earning capability, growth capability, and financial structure which five items including 27 variables, Factors can be collected by regression-analyze. Each year surely get different factors, and input these factors to backward- regression or stepwise- regression that can get some conclusions: 1、Input different factors by different years to backward- regression can different result. The last year can win the accurate rate to 93%. The second year can win the accurate rate to 86%. The third year can win the accurate rate to 90%. The fourth year can win the accurate rate to 83%. 2、Input different factors by different years to stepwise - regression can different result. The last year can win the accurate rate to 91%. The second year can win the accurate rate to 86%. The third year can win the accurate rate to 83%. The fourth year can win the accurate rate to 82%. 3、 Input factors by the sum of 2000 to 2003 years to stepwise-regression create a model and it would win the accurate rate to 91%, input factors of 2003 years replace factors of 2003 to 2003 years by the same model would wine the accurate to 84%. 4、 To test again factors of 2003 year by Discriminate Analysis can win the accurate rate 84%. The financial crisis predictive model will provide the information to investor and financial companies. But some company's financial account has been done with window dressing. If we went to protect investor and financier against investing crisis. We suggest that the administration must strict to manage the stock market